llLook.com

Oxidation of Cobalt Nanocrystals: Investigation of

the Role of Nanocrystallinity, Self-Ordering and

Nanocrystal Size Master of Science Thesis in the Master Degree Program,

Materials Chemistry and Nanotechnology

JOHANNA BERGSTRÖM

Department of Chemical and Biological Engineering

Division of Applied Chemistry

CHALMERS UNIVERSITY OF TECHNOLOGY

Göteborg, Sweden 2013

i

Oxidation of Cobalt Nanocrystals:

Investigation of the Role of Nanocrystallinity,

Self-Ordering and Nanocrystal Size

JOHANNA BERGSTRÖM

The thesis work was performed at Université Pierre et Marie Curie (UPMC) in

Paris, France in the Laboratoire des Matériaux Mésoscopiques et Nanométriques

and as a part of the SupraNano project group. The project group is directed by

Professor Marie-Paule Pileni.

Department of Chemical and Biological Engineering

CHALMERS UNIVERSITY OF TECHNOLOGY

Göteborg, Sweden 2013

ii

Oxidation of Cobalt Nanocrystals: Investigation of the Role of

Nanocrystallinity, Self-Ordering and Nanocrystal Size

JOHANNA BERGSTRÖM

© JOHANNA BERGSTRÖM 2013

Department of Chemical and Biological Engineering

Chalmers University of Technology

SE-412 96 Göteborg

Sweden

Telephone + 46 (0)31-772 1000

Cover: HRTEM images from polycrystalline fcc nanocrystals produced from organometallic

synthesis. One of the obtained structures after oxidation at 200˚C for 10 min, yolk/shell fcc-

Co/CoO particle showing the lattice planes for the fcc-Co core and the CoO shell. See page 14

for more information.

Göteborg, Sweden 2013

iii

Acknowledgements

I would like to thank Professor Marie-Paule Pileni for taking me into her group at Laboratoire des

Matériaux Mésoscopiques et Nanométriques, at UPMC in Paris. Thank you also to Zhijie Yang and

Dr. Jianhui Yang for valuable help and for answering all of my questions and thank you Professor

Krister Holmberg for putting me in contact with Professor Pileni and for valuable help from Sweden.

iv

Abstract The use of cobalt and cobalt oxide nanocrystals has proven promising in technical applications such as

information storage, catalysis and Li-ion batteries. Hollow cobalt oxide nanocrystals can be obtained

by using the nanoscale Kirkendall effect and the ability to tune the physical properties has made

research regarding size, shape, crystal structure and composition into a popular field. However, the

effect of the nanocrystallinity on ordered nanoparticles exposed to oxidation has not been considered

this extensively in previous research. The effect of nanocrystallinity has been carried out using four

different samples, all 8 nm before oxidation, of cobalt nanocrystals synthesized by two different

routes, the reverse micelle route for the samples small domain polycrystals and single domain hcp

nanocrystals. Synthesized by the second route, the organometallic synthesis, were the other two

samples, polycrystalline fcc nanocrystals and single domain ε nanocrystals. The samples were exposed

to an oxygen flow for 10 minutes during heating at 200°C or 260°C after which the nanocrystals were

characterized using TEM, HRTEM and ED. An investigation of how size affects the oxidation

behavior was carried out by using 4, 6 and 8 nm samples from the reverse micelle route. Poly- and

single domain crystal Co-nanoparticles showed different oxidation behavior. Single crystalline Co-

nanoparticles have a higher tendency to form hollow nanoparticles than polycrystalline Co-

nanoparticles. The difference in oxidation behavior is also clear between the two single crystalline and

between the two poly crystalline structures. Isolated nanocrystals have a higher probability of being

fully oxidized than ordered nanocrystals and the size of the nanocrystals as well as the oxidation

temperature has an effect on which oxide, the cubic CoO or the spinal Co3O4, is obtained after

oxidation.

Keywords: Co nanoparticles, nanocrystallinity, 2D ordering effect, nanocrystal size effect, core/shell

structures, hollow structures, yolk/shell structures.

v

Table of Contents 1. Introduction ..................................................................................................................................... 1

1.1. Purpose .................................................................................................................................... 1

1.2. Limitations............................................................................................................................... 2

1.3. Outline of the thesis ................................................................................................................. 2

2. Theory ............................................................................................................................................. 3

2.1. The Kirkendall effect ............................................................................................................... 3

2.2. Crystal structures of metallic cobalt ........................................................................................ 3

2.3. Structures of Cobalt oxides ..................................................................................................... 4

2.4. Evaporation-induced self-assembly ......................................................................................... 5

2.4.1. Colloidal stabilization by oleic acid ................................................................................ 5

3. Method............................................................................................................................................. 6

3.1. Experimental Procedure .......................................................................................................... 6

3.1.1. Sample preparation for TEM ........................................................................................... 7

3.1.2. Dry phase annealing ........................................................................................................ 7

3.1.3. Dry phase oxidation ......................................................................................................... 7

3.2. Characterization ....................................................................................................................... 8

3.2.1. Transmission Electron Microscopy (TEM) ..................................................................... 8

3.2.2. Electron diffraction (ED) ................................................................................................. 8

3.3. Statistics................................................................................................................................... 8

4. Results and Discussion .................................................................................................................... 9

4.1. Nanocrystallinity dependency ................................................................................................. 9

4.1.1. Polycrystalline nanocrystals produced from reversed micelles (Copoly-RM) ..................... 9

4.1.2. Single domain hcp nanocrystals produced by annealing Co nanoparticles from reverse micelle approach (CoHCP-RM) .......................................................................................... 11

4.1.3. Polycrystalline fcc nanocrystals produced from organometallic synthesis (Copoly-ORG) 14

4.1.4. Single domain ε phase Co nanocrystals produced from organometallic synthesis (Coε-ORG) ................................................................................................................................. 16

4.2. Size dependency .................................................................................................................... 18

4.2.1. Polycrystalline nanocrystals produced from reverse micelles (Copoly-RM) ..................... 18

4.2.2. Single domain hcp nanocrystals produced by annealing Co nanoparticles from reverse micelle approach (CoHCP-RM) .......................................................................................... 20

5. Conclusion ..................................................................................................................................... 22

5.1. Future work ........................................................................................................................... 23

1

1. Introduction

Metal oxides are commonly used today in a variety of applications from cathodes in batteries to

semiconductors in solar cells (1), (2). Hollow inorganic nanocrystals possess the interesting

phenomena of having an empty internal space which creates possibilities for usage in optical,

catalytically and magnetic devices such as nanoreactors and carriers for drug delivery (3). The ability

to tune physical properties of metal oxides nanoparticles has done the investigation regarding size,

shape, crystal structure and composition into a popular field (4), (3). However, not as much attention

has been focused on how the crystallinity of the nanoparticle, called the nanocrystallinity, affects the

physical properties (4), (5).

Magnetic transition metals, like cobalt (Co), are often air sensitive which makes them complicated to

work with. As a consequence of this, metal oxides are used in magnetic applications instead despite

the weaker magnetic properties, as they are stable in an air containing environment (6). Co and cobalt

oxides (CoO, Co3O4) are promising materials when it comes to usage in technical applications such as

catalysis, information storage and Li-ion batteries (7). CoO hollow nanocrystals can be successfully

synthesize by the Kirkendall effect starting from solid Co nanoparticles and exposing them to a flow

of oxygen (8). The Kirkendall phenomenon of diffusion arises during heating when two species in a

diffusion couple have different atomic diffusivities (9). As an effect, pores will form in the faster

diffusing material, because of a motion of the boundary layer, and can be supersaturated to form a

single interior void or a yolk-shell structure (8), (10). Diffusion is made easier by defects in the lattice,

like grain boundaries, since this enables atomic jumps. A hollow core formation can be prevented by

the produced phase being rich in defects which can absorb the vacancies without forming voids or by

the particle being too small to be stable as a hollow structure (11). The synthesis of hollow

nanocrystals through the Kirkendall effect provides a template free route which automatically

eliminates the sometimes difficult step of removing the template before obtaining the hollow particle

(3).

It has been shown that the nanocrystallinity of isolated 7 nm Cobalt nanocrystals does not play a role

for oxidation, as they are completely oxidized independently of their nanocrystallinity. As the particles

are assembled into a compact hexagonal network in a 2D lattice, the oxygen diffusion rate will be

slowed down, and the oxidation will be affected by the nanocrystallinity, leading to either core/shell

Co/CoO particles or hollow particles of CoO. (7) The effect of the nanocrystallinity on ordered

nanoparticles exposed to oxidation has not been considered this extensively in previous research.

1.1. Purpose

The purpose of the thesis is to investigate the control of nanocrystallinity through analysis of the

oxidation process of cobalt nanocrystals upon change in the crystalline structure of the nanoparticles

and their assemblies, and also to examine how nanocrystal size affects the oxidation process.

Investigations of four different cobalt samples have been made, fcc and hcp synthesized by reverse

micelle route, and fcc and ε formed by organometallic synthesis. The difference between the two fcc

samples, synthesized by different the routes, is that the reverse micelle route results in the formation of

smaller crystalline domains and thereby a more poorly crystallized structure, called small domain

polycrystal, than the organometallic synthesis. The two fcc structures therefor differ in

nanocrystallinity, as their domain size differs and the one with smaller domain size is closer to an

amorphous structure, even if they are both fcc and polycrystalline.

2

1.2. Limitations

Oxidation, annealing and characterization are a part of the thesis work however synthesis of the Co-

nanoparticles will not be included, but provided by the lab. For this reason, no synthetic routes will be

presented as the focus has been on the characterization and of the nanoparticles. Metallic cobalt has

three different crystal structures, which are all described more thoroughly in chapter 2.2, and the three

possible structures have been tested.

Nanocrystallinity dependency will be investigated using nanoparticles with a size of 8 nm and how

size affect the oxidation process will be examined for three different sizes, 4, 6 and 8 nm and only for

two of the samples, the crystal structures fcc and hcp synthesized by the reverse micelle route. The

reason why size dependency is not performed for all the samples is that there is not yet a successful

method to obtain a good enough size control for the smaller particles made by the organometallic

synthesis.

HRTEM has been performed on the samples before oxidation and for the samples oxidized at 200°C

for 10 minutes but not for the samples oxidized at 260°C for the same time. The reason for this is that

the HRTEM is a time consuming process and the instrument can only be run by trained people. In the

end the time was not enough and the 200°C oxidation treatment was prioritized.

1.3. Outline of the thesis

The report is divided into five chapters, chapter 2, Theory, is presenting the necessary background

information needed to understand the experimental methods and the oxidation behavior. It deals with

the diffusion behavior described by the Kirkendall effect, the different structures of cobalt and cobalt

oxides, the self-assembly and the colloidal stabilization necessary to prevent the particles from

coalescence. Chapter 3 deals with the methods which in more elaborated terms mean the experimental

procedure and the description of the methods for characterization, Transmission electron microscopy

(TEM), High Resolution TEM (HRTEM) and electron diffraction (ED). Chapter 4, Results and

Discussion, is divided into two parts. The first part handles the effect of nanocrystallinity and self-

ordering while the second part presents the results regarding the size effect. The thesis is ended with

chapter 5, conclusions, concluding the work and suggesting further areas of investigations.

3

2. Theory

The theory chapter contains background about the Kirkendall effect used to create hollow CoO or

core/shell Co/CoO, information about the three different crystal structures of metallic cobalt and

cobalt oxide, and the theory behind the self-assembly used to deposit samples on TEM grids for

characterization.

2.1. The Kirkendall effect

During solid state reactions, diffusivity is a kinetically determining factor. Charge and size of the

particles taking place in the reaction control the diffusivity. If two particles with different diffusivity

react with each other the reaction will be faster in one direction than the other. This is known as the

Kirkendall effect and results in a motion of the boundary layer that can be used to produce hollow

nanoparticles. Oxygen is a large and charged ion and even though the metal ions in the lattice are also

charged, the smaller size of the metal ion compared to the oxygen is what most often make up for the

diffusion difference. This causes the metal ions to migrate outwards rather than the oxygen ions to

migrate into the metal lattice, the created vacancies migrate inwards and can form a single interior

cavity. (10)

Figure 1. The nanoscale Kirkendall effect. The metal nanocrystal is reacted with oxygen and because of the faster

outward diffusion of metal than the inward diffusion of oxygen voids are formed in the lattice. The voids are saturated

in the center leading to a hollow nanocrystal.

The Kirkendall effect is known for causing weakness within a bulk material as it is the reason for

internal void formation and thereby weakening of the material, but on the nanometer scale it can be

used to create hollow nanoparticles instead, because of super saturation of vacancies into a hollow

core (3), (11)., see Figure 1. The nanoscale Kirkendall effect is more complex than on a macroscale

since the nanoscale system include factors like a shorter diffusion length, interface stress and

inhomogeneous coating, all of which are less significant on a macroscale. Diffusion is made easier by

defects in the lattice, like grain boundaries, since this enables atomic jumps. A hollow core formation

can be prevented the produced phase being rich in defects which can absorb the vacancies without

forming voids or by the particle being too small to be stable as a hollow structure. (11)

2.2. Crystal structures of metallic cobalt

Cobalt is a transition metal and that can be found between iron and nickel in the periodic system of

elements and it is naturally occurring in the crust of the earth (2). Metallic cobalt has three possible

crystal structures, the hexagonal closed-packed (hcp), face-centered cubic (fcc) and epsilon phase (ε)

(12). The three phases are shown in in Figure 2 A-C. At normal pressure, two crystal structures are

stable for bulk Co, the hcp below 425°C and the fcc at higher temperatures (2), (13). Because of a low

free energy exchange, the change into the more stable hcp in room temperature is very slow (2).

4

Metastable phases such as ε-Co nanocrystals are possible to form in solution phase chemical synthesis

because it is as a general rule not controlled by thermodynamics.

The hcp structure is constructed of layers in an ABABAB-stacking sequence, this leaves small

channels in the structure between the Co-atoms. The fcc structure has an ABCABC-stacking sequence,

making it cubic. The ε-phase is composed of two different geometrical arrangements, having the cubic

structure of β-manganese, some distances are shorter than others and the purple and blue atoms in

Figure 2 C have their neighboring atoms arranged differently.

Figure 2. Sketch view of different bulk cobalt structures: (A) hcp or α, (B) fcc or β, (C) ε, (including a detail of the

different cobalt position). [From reference (12), Reprinted with permission from American Institute of Physics. ©

2010 American Institute of Physics.]

During annealing, heat treatment, of a Co sample at a temperature below 425 ᵒC the metal ions in the

lattice becomes more mobile than in room temperature. This causes the ε or fcc lattice to transform

into the more stable hcp structure (6), (14). The mechanics behind the change is due to dislocation

movements of the octahedral planes of the cubic fcc or ε lattice (2).

Single domain crystals are built up by one single crystal, while polycrystals are built up by two or

more fused crystals which mean that grain boundaries are present within a polycrystal. Magnetic

properties of nanocrystals depend strongly on both particle size and the crystal structure. The complex

cubic structure of the ε-Co nanocrystal is a soft magnetic material meaning the coercivity is low and

the required force to demagnetize the material after it has been magnetized is low (6). The hcp is

ferromagnetic at all temperatures, and the magnetic behavior remains after the source of the alignment

of the electric dipoles has been removed. However the fcc becomes paramagnetic at 1121 ± 3ᵒC,

resulting in the consequence that the electric dipoles orient randomly after the removal of the magnetic

field above the given temperature. Single crystalline hcp is magnetically anisotropic and will be more

easily magnetized in on direction than the other, while the cubic structures have an isotropic magnetic

behavior. The magnetic behavior of polycrystalline particles depend both of the purity of the metal and

the thermal history of the material (2).

2.3. Structures of Cobalt oxides

Cobalt oxide complexes have been traditionally used as dying agents in the ceramic and glass

production, it adapts different color depending on the metal ion binding together with the cobalt oxide.

The density for the oxides is lower than for metallic cobalt leading to a growth in size of cobalt

nanoparticles during oxidation (2).

5

Cobalt has two readily available oxidation states, Co2+

and Co 3+

, and cobalt oxides exist in two main

structures. The structurally most simple one is cobalt(II)oxide (CoO) which has a rock salt structure,

meaning one Co2+

is octahedrally coordinated to the oxygen atoms in the lattice, with a = 0.425 nm.

The structure is stable above 900 ᵒC and is antiferromagnetic at normal temperatures. CoO has a

density of 6.44 g/cm3

at 20ᵒC (15).The second main structure is cobalt(II,III)oxide (Co3O4), it is

divalent and has a spinel structure with tetrahedrally coordinated Co2+

and octahedrally coordinated

Co3+

, with a = 0.807 nm. It is the thermodynamically stable form of cobalt oxide under ambient

temperature and pressure (1), (2), (16). The larger separation between the lattice fringes means that

Co3O4 has a lower density than CoO with a value of 6.11g/cm3 at 20ᵒC

(15).

Cobalt oxide exists in a third composition as well, the cobalt(III)structure (Co2O3), it is unstable but

impure forms have been prepared. There is never a transition from Co3O4 into Co2O3, but the

chemistry of Co(III) is instead dominated by formation of other chemical complexes (1), (2), (16).

The oxides can exist as either poly- or single crystalline structures which will affect their properties

(2). Polycrystals can be used in catalysts or as nanoscale reactors as the grain boundaries enable a

possible route for diffusivity of small particles (8).

2.4. Evaporation-induced self-assembly

The compact hexagonal 2D networks studied in the project are self-assembled through evaporation-

induced self-assembly. A stable colloidal dispersion of Co nanocrystals is deposited on a surface and

the solvent is allowed to evaporate. The ordering takes place in the meniscus, which is the curved

surface formed at the top of the liquid, and because of the small volume of it, the nanocrystals are

forced together. The two most important repulsive interactions of the system are the steric hindrance

from the surfactants covering the surface of each nanocrystal due to reduced entropy and the osmotic

flow of solvent towards the compressed region, and the columbic repulsions due to the same signs of

the surface charges. The two forces together maximize the distance between the nanocrystals which

creates an ordered compact network. (10)

2.4.1. Colloidal stabilization by oleic acid

All samples presented in the report are coated with oleic acid, C18H34O2, see Figure 3. Steric

stabilization is necessary to avoid aggregation in the colloidal dispersion and the carboxylic group in

oleic acid binds strongly to the particle surface and forces the long hydrocarbon chain out into the non-

polar solution (17). The long hydrocarbon chain probably also has an effect on the oxygen diffusion as

is will work as a barrier for the nanocrystal and a protection against oxidation as well as aggregation.

Oleic acid has a melting point at 13 – 14ᵒC and a boiling point at 194 – 195ᵒC (18).

Figure 3. Chemical structure of oleic acid.

H3C (CH2)7 CH CH (CH2)7 C OH

O

6

3. Method

The chapter describes the experimental procedure, the characterization methods and statistics. The

experimental procedure starts with an overview before going into the different steps, after which

characterization methods are presented one by one before a short section on how statistical certainty

was achieved during analysis.

3.1. Experimental Procedure

Investigation of four different nanocrystalline cobalt samples have been performed, fcc synthesized by

reverse micelles, which is a polycrystal with small crystalline domains (sample 1), hcp synthesized by

reversed micelles, which has a single crystalline structure (sample 2), fcc formed by organometallic

synthesis, which is a polycrystal (sample 3) and ε formed by organometallic synthesis which has a

single crystalline structure (sample 4). The four investigated samples are summarized in Table 1

below. The difference between the two fcc samples, synthesized by the different routes, is that the

reverse micelle route results in the formation of smaller crystalline domains and thereby a more poorly

crystallized structure, called small domain polycrystal, than the organometallic synthesis. The two fcc

structures therefor differ in nanocrystallinity, as their domain size differs and the one with smaller

domain size is closer to an amorphous structure, even if they both have the fcc structure and are

polycrystalline.

Table 1. Summary of investigated samples, their structure, nanocrystallinity and the synthetic route for formation.

Sample no. Structure Nanocrystallinity Synthetic route Size (nm) Abbreviation

1 fcc Small domain polycrystal Reverse micelle 4, 6, 8 Copoly-RM

2 hcp Single domain Reverse micelle 4, 6, 8 CoHCP-RM

3 fcc Polycrystal Organometallic

synthesis

8 Copoly-ORG

4 ε Single domain Organometallic

synthesis

8 Coε-ORG

The stabilizer of the colloidal dispersion was in all four cases oleic acid, which is described in chapter

2.4.1. Comparison of the nanocrystallinity effect was performed with nanoparticles in the size of 8 nm

and the size effect has been examined using three different sizes, 4, 6 and 8 nm, of fcc and hcp

synthesized by the reverse micelle route. Characterization of the nanoparticles was executed using

TEM, HRTEM and electron diffraction.

Sample 1, 3 and 4 were synthesized in a glove box with nitrogen atmosphere to avoid oxidation of the

nanoparticles. To achieve the hcp structure of sample 2, sample 1 was dry-phase annealed in 250 ᵒC

under argon flow for 60 minutes. All samples were oxidized in 200˚C or 260˚C for ten minutes and

characterized with TEM and ED both before and after oxidations to investigate how the oxidation

procedure had been affected by the nanocrystallinity, the self-ordering and the nanocrystal size.

HRTEM was performed on all samples after oxidation to receive more information about the oxidation

behavior. A summary of the experimental procedure is presented in the flowchart in Figure 4.

7

Figure 4. Flow chart of experimental procedure for investigation of how nanocrystallinity, self-ordering and

nanocrystal size affects the oxidation of Co nanocrystals.

3.1.1. Sample preparation for TEM

The sample was prepared in a glove box with inert nitrogen atmosphere. The sample, containing Co-

nanoparticles coated with oleic acid in hexane, was deposited on the TEM grid (copper grid coated

with amorphous carbon film) and left for two minutes allowing the solvent to evaporate causing the

Co-nanoparticles to self-assemble into a monolayer. The grid was placed in the sample loader and

inserted into the TEM for characterization.

3.1.2. Dry phase annealing

The samples were annealed through dry phase annealing, meaning that the particles were already

deposited and ordered into a 2D lattice on the TEM grid when exposed to the heat treatment. The grid

with the applied sample was placed in a tube and loaded on a modified Schlenk line, which included a

high vacuum pump, a source of inert argon gas and a source of oxygen gas. The sample was exposed

to vacuum then Ar flow in three rounds to ensure that all air had been removed from the system. The

heater was adjusted to 250 ᵒC and the annealing took place under argon flux for 60 minutes, after

which the heater was removed to stop the annealing. The sample was left to cool to room temperature

by the argon flow (10 min).

3.1.3. Dry phase oxidation

The grid with sample on it was placed in a tube and loaded on the modified Schlenk line, the same

used during annealing, and treated with O2 (200˚C, 10 min or 260˚C, 10 min). After oxygen treatment

the heater was removed, the oxygen flow was replaced by argon flow to stop the oxidation

immediately and the sample was allowed to cool to room temperature by the argon flow (10 min).

8

3.2. Characterization

Methods used for characterization of the Co-nanoparticles, both before and after the two oxidations,

are TEM, HRTEM and ED. The methods are described in the two following subchapters.

3.2.1. Transmission Electron Microscopy (TEM)

The analysis is performed in a vacuum chamber. A high energy beam of electrons is directed towards

a thin sample of the particles assembled on a copper grid coated with an amorphous carbon film. The

electrons that are transferred through the sample are detected and form a two dimensional projection of

the sample. The instrument can be chosen to detect either the direct beam (bright field image) or the

diffracted beam (dark field image) (19). TEM can be used as a high resolution instrument to detect the

size and shape of the particles, but not to determine the 3D structure of the lattice.

TEM is available in different modes, depending on what instrument is used. In High Resolution mode

(HRTEM), the electrons are accelerated with a higher potential field, giving them a shorter wavelength

and higher energy which increase the resolution of the obtained image (19). This way the lattice planes

of a nanocrystal oriented in the right direction can be seen.

TEM was performed using a JEOL 1011 microscope at 100 kV, and HRTEM was performed using a

JEOL 2010 microscope at 200 kV.

3.2.2. Electron diffraction (ED)

Electron diffraction (ED) is available in the TEM instrument and can be used to determine the crystal

structures of the nanocrystals. Electrons are fired at the sample and the resulting interference pattern is

observed as the incoming primary electrons interact with the secondary electrons reflected off the

lattice planes. Electron diffraction can be used to calculate the interplanar distance between lattice

planes and can thereby be used to determine the crystal structure (20). The diameter of the resulting

rings in the ED patterns are measures and the inverse value of the radius can be compared to

theoretical values to determine, confirm or deny the crystal structure.

3.3. Statistics

To measure size of the full diameter or the internal structure and achieve histograms and size

distributions of the nanocrystals both before and after oxidation, at least 500 nanocrystals were

measured on the TEM images for each sample, using the free software ImageJ. The histograms were

prepared using Origin8. In the cases where the percentage of hollow, core/shell or fully oxidized

particles are presented, at least 500 nanocrystals were investigated, using the TEM images.

9

4. Results and Discussion

The chapter is divided into two main subchapters, the first one, chapter 4.1, concerns the investigation

on how the nanocrystallinity affects the oxidation behavior and the second one, chapter 4.2, is about

how the size of the nanoparticle affects the oxidation behavior. Both subchapters deal with the

question on how the self-ordering affects the end results. The nanocrystallinity dependency is

examined using 8 nm particles for all four different samples while the size effect is evaluated using 4,

6 and 8 nm particles of Copoly-RM and CoHCP-RM. The expected result before the start of the investigation

was to find a change within the oxidation of the various single domains and between the single- and

polycrystals.

4.1. Nanocrystallinity dependency

The results regarding the 8 nm particles for each of the four different nanocrystallinities are presented

separately for each sample and discussed in the text. The samples are all presented before oxidation,

after oxidation at 200°C for 10 minutes and after oxidation at 260°C for 10 minutes. After the last

sample a more general discussion is made, taking all the results into account and trying to answer how

nanocrystallinity and self-ordering affects the oxidation behavior of cobalt nanocrystals.

4.1.1. Polycrystalline nanocrystals produced from reversed micelles (Copoly-RM)

The polycrystalline nanocrystals synthetized by the reverse micelle route (Copoly-RM) are characterized

by very small fcc domains that indicate a very low crystallinity. No evidence can be given to prove

that some domains are not amorphous. In Figure 5a, a Copoly-RM nanoparticle is visualized in a HRTEM

image. No crystalline domains are clearly seen as they are too small to appear as an ordered structure

in the images, this indicates that the crystallinity is not well defined.

Figure 5. HRTEM images from polycrystalline nanocrystals produced from reversed micelles (Copoly-RM) 8 nm.

(a) Native nanoparticle before oxidation, no lattice planes can be perceived which indicates low crystallinity, (b) After

oxidation at 200˚C for 10 min. Core/shell Copoly-RM/CoO nanoparticle showing the lattice plane (200) for CoO in the

shell.

Figure 6a and 6b show the TEM images at various scales of the nanocrystals in their native state

before oxidation. The observed zone is well ordered in a compact hexagonal network with some

isolated nanocrystals, giving the opportunity to study how the 2D ordering affects the oxidation. From

the histogram (inset Figure 6a) the nanocrystal size and its distribution are 8.3 ± 0.77 nm and 9.3%

respectively. The electron diffraction (ED) pattern (Figure 6c) shows only one clear ring

corresponding to the fcc lattice plane (111). This confirms the information from the HRTEM image of

the same sample in Figure 5a in which the crystallinity of the nanoparticles is not well defined.

10

Figure 6. Polycrystalline nanocrystals produced from reversed micelles (Copoly-RM), 8 nm. (a, b) TEM images before

oxidation, (d, e) after oxidation at 200˚C for 10 min, (g, h) after oxidation at 260˚C for 10 min. (c, f, i) Corresponding

ED pattern before and after the two oxidation temperatures. The insets in (a, g) are histograms displaying the size

distributions of the nanoparticle diameter, in (d) the histogram displays both the particle diameter and the diameter

of the core in the core/shell nanoparticle.

The TEM images of the sample submitted to oxidation at 200°C for 10 minutes reveals formation of

both core/shell (Copoly-RM/CoO) and CoO nanocrystals (Figure 6d and 6e). The average diameter of

nanocrystals is 10.6 ± 0.98 nm, with a size distribution of 9.2%, which is almost equal to the

distribution observed before oxidation (9.3%) even if the nanoparticles have grown. The average

diameter of the core proved to be 3.9 ± 1.07 nm, with a size distribution of 27.8%. The calculated

thickness of the oxide shell is thereby 3.4 nm, however, there is large size distribution of the core

making the calculations uncertain. From HRTEM image (Figure 5b) it can be seen that the crystalline

structure of Co core remains similar as that observed before oxidation while the oxidized shell shows a

polycrystalline structure with a lattice plane corresponding to the CoO (200). A particle in more highly

ordered areas of the lattice has the tendency to become core/shell during oxidation, but along the edges

of the monolayer and in the regions where the nanocrystals are not well ordered, most of the

nanocrystals are fully oxidized.

In Figure 7a, a small assembly of nanoparticles is visualized, it is clearly seen that the nanoparticles in

the middle of the assembly are core/shell while the edges of the assembly consists of nanoparticles

with a smaller core or are fully oxidized. It can be noticed that isolated nanoparticles are mostly fully

oxidized, but some size dependency can be seen among them, where larger particles received a

core/shell structure while smaller particles were fully oxidized. In Figure 7b two isolated particles

show a core/shell structure with a very small core. A general feature is that the larger the nanoparticle

size is, the larger the core becomes in the core/shell structure for both ordered and isolated structures.

The size effect of the Copoly-RM has been further investigated in chapter 4.2.1. The ED pattern (Figure

6f) displays rings for both CoO and fcc-Co supporting the information given by the HRTEM image

11

that a core shell structure with both the fcc-Co and CoO is present. However, only the outer rings of

the fcc lattice can be seen but as fcc (111) and (220) is close in d-spacing to CoO (200) and (311) they

can be hidden behind each other, or the rings for the fcc lattice can also be unclear because of the

nearly amorphous structure. The separation between the lattice planes is larger for CoO than fcc-Co,

and this change in density explains the growth during oxidation (2), (15).

Figure 7. Polycrystalline nanocrystals produced from reversed micelles (Copoly-RM), 8 nm, after oxidation at 200˚C for

10 min. (a) TEM images of small assembly of nanoparticles. (b) TEM image of isolated particles. The inset is the

corresponding low magnification TEM where the enlarged image is selected from the white rectangle.

When the oxidation process takes place at 260°C instead of 200°C, keeping the same exposure time

(10 min) some core/shell structure is still present, but most of the nanoparticles have become fully

oxidized (Figure 6g and 6h). The size of the particles has increased to 11.7 ± 1.09 nm but keeping the

distribution of 9.3%. According to the corresponding ED pattern (Figure 6i) the oxide has obtained the

spinal structure, Co3O4, in contradiction with the oxidation at 200°C where the nanoparticles received

the cubic CoO structure. The space between the nanoparticles decreases on increasing the temperature

and consequently the tendency of nanocrystals to coalescence increases. This can be due to the faster

oxidation at 260˚C compared to 200˚C or a consequence of the smaller space between the lattice

planes and thereby lower density of the Co3O4 compared to the CoO (15).

In conclusion, Copoly-RM nanocrystals produced from the reverse micelle route is highly polycrystalline

and thereby presents many grain boundaries for possible oxygen diffusion. As the oxygen diffusion

becomes fast, the inward diffusion can compete with the outward diffusion of Co atoms and no

internal void is formed. The oxidation thereby results in a core/shell particle. In the case that internal

pores are formed they can be annihilated in the high concentration of grain boundaries without

forming an internal void (11). In the less ordered areas the Copoly-RM nanoparticles are exposed to more

oxygen compared to the nanoparticles in the more highly ordered areas, as the nanoparticles are not

being protected from the oxygen by the neighboring particles on the sides, resulting in fully oxidized

nanoparticles instead of the core/shell particle. These data are in agreement compared to those already

obtained with 7nm Copoly-RM nanoparticles performed earlier by the same group (7). The higher

oxidation temperature of 260˚C speeds up the oxidation, which results in a higher concentration of

fully oxidized particles compared to after the lower oxidation temperature 200˚C.

4.1.2. Single domain hcp nanocrystals produced by annealing Co nanoparticles from reverse micelle approach (CoHCP-RM)

Single domain crystalline hcp (CoHCP-RM) are produced by annealing Copoly-RM nanoparticles. Figure 8a

and 8b show TEM images with well-defined nanocrystals with an average size and distribution of 8.4

± 0.85 nm and 10.1% respectively. The size is slightly larger than what was observed for Copoly-RM

12

nanoparticles (8.3 ± 0.77 nm and 9.3%), but the small change is within the margin of error of the

measurements. Figure 9a shows the HRTEM images for CoHCP-RM before oxidation, the ordered lattice

planes correspond to the hcp-Co (002) and the nanocrystal is built up by one single crystalline domain.

No sign of oxidation from the annealing treatment can be seen as this would have shown on the

HRTEM image and also have generated a ring, characteristic for CoO, inside the inner triplet rings in

the ED pattern (Figure 8c).

Figure 8. Single domain hcp nanocrystals produced by annealing Co nanoparticles from reverse micelle approach

(CoHCP-RM) 8 nm. (a, b) TEM images before oxidation, (d, e) after oxidation at 200˚C for 10 min, (g, h) after oxidation

at 260˚C for 10 min. (c, f, i) Corresponding ED pattern before and after the two oxidation temperatures. Tint and Text

marked on the ED pattern in (c) and (f) refer respectively to the internal triplet rings [(100), (101), (002)] and external

triplet rings [(110), (103), (112)], of the hcp lattice. The insets in (a, g) are histograms displaying the size distributions

of the nanoparticle diameter, in (d) the histogram inset displays both the particle diameter and the diameter of the

core in the core/shell nanoparticle.

The CoHCP-RM nanocrystals are submitted to same treatment as the previously described nanoparticles

(oxidation at 200°C; 10mn). For any ordered nanocrystal a core/shell CoHCP-RM/CoO structure is

observed (Figure 8d and 8e), but for isolated nanocrystals either core/shell or fully oxidized

nanocrystals are produced. The thickness of the shell is thinner when the nanocrystals are ordered

compared to when they are isolated, this behavior is visualized in Figure 8e with the single isolated

particle having a smaller core, and thereby a thicker shell, than the neighboring ordered particles.

Furthermore, the shell is thinner for the largest nanocrystals compared to smaller one when both are

ordered in compact hexagonal network. The nanocrystals have a mean diameter of 9.4 ± 0.88 nm and a

size distribution of 9.4%, the average diameter of the core for ordered nanoparticles is 5.8 ± 0.91 and

has a distribution of 15.7% this gives the average shell thickness of 1.8nm. The HRTEM image

(Figure 9b) shows that the large core keeps its former crystalline structure. It has a regular pattern of

lattice planes that correspond to the hcp-Co (002) plane. The oxide shell shows two different distances

for the CoO lattice, 2.13 Å corresponding to the (200) plane and 2.46 Å corresponding to the (111).

The corresponding ED pattern (Figure 8f) illustrates a mixture of both hcp-Co and CoO which

supports the information from the TEM and HRTEM pictures of a core/shell CoHCP-RM/CoO structure.

These data are reproducible and surprising if compared to a previous study performed by the same

group with 7 nm CoHCP-RM nanoparticles instead of 8 nm. For the 7 nm nanoparticles, hollow CoO

13

single domain nanocrystals were produced (7). The effect of size on the oxidation behavior of CoHCP-

RM has been investigated further in chapter 4.2.2.

Figure 9. HRTEM images from single domain hcp nanocrystals produced by annealing Co nanoparticles from reverse

micelle approach (CoHCP-RM) 8nm. (a) Before oxidation, single domain nanoparticle showing the lattice planes for the

hcp (002) plane. (b) After oxidation at 200˚C for 10 min. Core/shell Co/CoO nanoparticle showing the lattice planes

for the hcp core (002) and the CoO shell (111) and (200).

When the temperature is increased to 260°C instead of 200°C, keeping the same exposure time, the

nanoparticles tend to coalesce (Figure 8g and 8h). However, it is shown that the nanoparticles present

hollow structure with a thick oxide shell and an internal void. An increase in the size can be seen after

oxidation at 260˚C compared to after 200˚C as the particles are now 11.4 ± 1.33 nm with a distribution

of 11.7%. The ED pattern (Figure 8i) shows formation of the spinal structure Co3O4 instead of CoO.

The difference in sizes observed between the sample oxidized in 200°C and in 260°C is explained by

the fact that Co3O4 has lower density compared to CoO and also that a more rapid oxidation and

consequently a faster oxygen diffusion takes place after increasing the temperature. The creation of an

internal void is also a factor that leads to a larger increase in the outer diameter

During the annealing process it is likely that some of the oleic acid used as a coating agent is broken

down, but as no fusion of the Co can be seen after the annealing, some coating must still be intact

surrounding the nanoparticles and making them avoid aggregation. The partly broken oleic acid can be

the reason that CoHCP-RM oxidized at 260°C for 10 min coalescence more than Copoly-RM after the same

oxidation treatment. However, the formation of an internal void for CoHCP-RM and thereby a larger

increase in the size is also a reason for enhanced coalescence. It can clearly be seen that 200°C for 10

minutes does not produce a fast enough outward movement of cobalt compared to the inward

movement of oxygen to create hollow nanoparticle for the 8 nm structure.

If CoHCP-RM is compared to Copoly-RM (chapter 4.1) for the 8 nm nanoparticles, they differ in their

oxidation behavior. Both samples form core/shell structures after oxidation at 200°C for 10 min but

the size of the cores are larger for the hcp single crystalline structure than the polycrystalline

nanoparticle, and the size change during oxidation is smaller for the single domain nanocrystal. The

difference can clearly be seen when comparing the thickness of the oxide shell. For Copoly-RM the shell

thickness is 3.9 nm while CoHCP-RM displays a thickness of 1.8 nm. The difference can be an indication

that oxygen atoms have a slower diffusion path in to the hcp single crystal lattice than the

polycrystalline lattice. Because of the low crystallinity many grain boundaries are present in the

polycrystalline lattice while no grain boundaries are present in the single crystalline hcp lattice. The

grain boundaries work as a fast diffusion path making the oxidation of easier for Copoly-RM than for

CoHCP-RM. A faster oxidation means a smaller core for the structure after oxidation and thereby a

thicker oxide shell.

14

4.1.3. Polycrystalline fcc nanocrystals produced from organometallic synthesis (Copoly-ORG)

The organometallic method is used to synthesize polycrystalline fcc-Co (Copoly-ORG) with larger

crystalline domains compared to those produced within the Copoly-RM nanocrystals, grain boundaries

are thereby present, but not in as high concentration as for the sample prepared by reversed micelles.

HRTEM images (Figure 10a) shows some grain boundaries, verifying the poly crystallinity of the

sample, but if compared to Copoly-RM in Figure 5a, Copoly-ORG clearly has larger crystalline domains. The

Copoly-ORG nanocrystals deposited on TEM grid shows mainly 2D super lattices and the formation of

multilayers (Figure 11a and 11b) and very few isolated nanocrystals are detected. From histogram

shown in the inset in Figure 7a, the average diameter of Copoly-ORG nanocrystals before oxidation is 8.1

± 0.67 nm with a 7.1% size distribution. The ED pattern (Figure 11c) clearly shows presence of an fcc

structure.

Figure 10. HRTEM images from polycrystalline fcc nanocrystals produced from organometallic synthesis (Copoly-ORG)

8nm. (a) Before oxidation, lattice planes nanoparticle showing lattice planes for fcc-Co. (b) One of the obtained

structures after oxidation at 200˚C for 10 min, Core/shell Co/CoO particle showing the lattice planes for the fcc-Co

core and the CoO shell, (c) the other obtained structure after oxidation at 200˚C for 10 min, hollow particle showing

the lattice planes of the CoO shell.

After submitting Copoly-ORG nanocrystals deposited on a TEM grid to oxidation (200°C; 10 min) both

yolk/shell (Co-fcc/CoO), and hollow CoO nanoparticles were produced (Figure 11d and 10e). With

the term yolk/shell means that the core is isolated from the shell and that the void is saturated between

the core and the shell unlike a core/shell particle where the metal core and the oxide shell are in

contact with each other. The average size of nanocrystals markedly increased compared to Copoly-ORG

nanocrystals before oxidation. It is found equal to 10.7 ± 0.80 nm with a size distribution of 7.5%

(inset Figure 11d). The HRTEM image shows a yolk/shell nanocrystal (Figure 10b) with the Co core

isolated from the CoO shell by voids. The core shows a single domain crystalline structure with

regular lattice planes corresponding to the fcc-Co (111) plane. The shell shows planes with a distance

of 2.13 Å corresponding to the CoO (200) plane. It is not possible to know if CoO shell is single

domain or polycrystals because the orientation of the shell could be such that only one side of the shell

shows the lattice planes, and none of the two possibilities could be confirmed or ruled out by a careful

HRTEM study. Figure 10c shows the formation of a hollow nanocrystal, after the same oxidation

treatment, with a shell characterized by lattice planes with a distance of 2.46 Å corresponding to the

CoO (111) plane. The oxide shell of the hollow particle is polycrystalline, as the directions of the

(111) planes differ in different areas. From TEM images, under investigation of 500 nanoparticles in

monolayers 62 % of nanocrystals are characterized by a yolk/shell structure and 38 % by hollow

nanoparticles. The shape of the particles is more irregular after oxidation than before. The

corresponding ED pattern (Figure 11f) supports the fact that both CoO and fcc-Co are present after

oxidation.

15

Figure 11. Polycrystalline fcc nanocrystals produced from organometallic synthesis (Copoly-ORG) 8 nm. (a, b) TEM

images before oxidation, (d, e) after oxidation at 200˚C for 10 min, (g, h) after oxidation at 260˚C for 10 min. (c, f, i)

Corresponding ED pattern before and after the two oxidation temperatures. The insets in (a, d, g) are histograms

displaying the size distribution at the different times.

The data after oxidation at 200°C for 10 min are rather surprising if compared to those obtained with

same size of Co nanocrystals produced from reverse micelles (Copoly-RM) with those produced from

inorganic method (Copoly-ORG). In both cases fcc domains are present before oxidation, however the

Copoly-RM nanocrystals (Figure 5a) are characterized by small well-defined domains whereas with

Copoly-ORG nanocrystals the fcc domains are larger (Figure 10a). Furthermore, it is observed that the

size of crystalline fcc domains in the Copoly-ORG nanocrystals before oxidation changes from one

nanocrystal to another. Such change within a given synthesis mode could explain the change in

morphology obtained after oxidation and explain the formation of the two different structures, the

yolk/shell and the hollow nanocrystals. With this in mind, the different structures obtained for Copoly-

RM and Copoly-ORG nanocrystals respectively after oxidation at 200°C for 10 min could also be

understood.

From a previous study performed by the same group (7) with 7nm Copoly-RM and CoHCP-RM nanocrystals

we known that the relative diffusion of oxygen and Co in the nanocrystals markedly changes between

hcp single domain and polycrystalline fcc phase. Here it becomes obvious that even though the

structure remains similar (fcc), the size of the domains significantly change the diffusion process. We

could assume, in first approximation, that with small fcc single domain the oxygen diffuse into the Co

lattice keeping the integrity of Co core instead of the Co diffusing outwards and creating voids within

the nanoparticle after oxidation treatment.

After oxidation at 260 ᵒC for 10 minutes (Figure 11g and 11h) the majority of nanoparticles show a

hollow structure, with some elements of yolk/shell particle with a small core. The assembly on the

TEM grid was uneven and no conclusions about self-assembly could be made. The average

nanocrystal (11.4 ± 0.92 nm with a distribution of 8.1%) increases compare that what observed under

16

oxidation at 200°C to the size after oxidation at 260°C, as expected. The void has a slight square shape

compared to the circular nanoparticles before oxidation and this behavior can also be seen when

viewing the particles after oxidation at 200°C. The corresponding ED pattern (Figure 11i) shows the

presence of Co3O4 which means the elevated temperature causes the structure to transform to a spinel

structure instead of the rock salt structure of CoO formed after oxidation at 200°C.

4.1.4. Single domain ε phase Co nanocrystals produced from organometallic synthesis (Coε-ORG)

Figure 12a and 12b show single domain ε phase Co nanocrystals (Coε-ORG) produced from

organometallic synthesis deposited on a TEM grid with an average size of 8.1 ± 0.58 nm and with a

size distribution of 7.1% (inset Figure 12a). The large and lighter monolayers are seen with some

small areas being double or triple layers. From the HRTEM image in Figure 13a, it is clear that the

nanoparticle consists of one single crystalline domain displaying the ε-Co (221) lattice plane. The

corresponding ED pattern presents a triplet, characteristic for the ε-Co (221), (310) and (311) lattice

planes.

Figure 12. Single domain ε phase Co nanocrystals (Coε-ORG) produced from organometallic synthesis, 8 nm. (a, b)

TEM images before oxidation, (d, e) after oxidation at 200˚C for 10 min, (g, h) after oxidation at 260˚C for 10 min. (c,

f, i) Corresponding ED pattern before and after the two oxidation temperatures. T marked on the ED pattern in (c)

refers to the triplet rings [(221), (310), and (311)] of the ε lattice. The inset in (a) is a histograms displaying the particle

diameter and the histogram insets in (d, g) displays the particle diameter and the diameter of the void.

Like the others Co nanocrystals, the sample was submitted to an exposure of oxygen in the same

experimental conditions as described above (200˚C; 10 min). The TEM images obtained after the

oxidation process shows formation of hollow nanocrystals (Figure 12d and 12e). The mean diameter

of the ε-Co nanoparticles increases compared to that obtained before oxidation and is found equal to

9.3 ± 0.75 nm with an 8.0% distribution. The increase in size is probably due to the outwards

migration of Co-particles during oxidation which is what also creates the central void, with an average

diameter of 3.1 ± 0.62 nm and a distribution of 20.2%. The two measurements reveal that the shell

thickness is 3.1 nm. The hollow CoO exist in a large quantity and only a few single core/shell

17

structures are obtained within the nanocrystal monolayer and. Some areas show particles with larger

voids within the more ordered areas of the lattice compared to particles in the less ordered areas.

Because of the lower order, the sample is in contact with a higher concentration of oxygen which

makes the inward diffusion of oxygen faster and therefor the internal void, caused by outward

diffusion of Co atoms, smaller. The HRTEM image (Figure 13b) shows that the shape of the Coε-ORG

nanoparticles has become more irregular after oxidation at 200°C for 10 min compared to before

oxidation (Figure 13a). The oxide shell shows regular lattice planes with a distance of 2.46 Å that

correspond to the CoO (111) planes with two different directions. This means that the CoO shell is

polycrystalline with two different crystalline domains. The corresponding image from the ED pattern

(Figure 12f) shows a clear pattern from CoO, which proves the results visualized on the TEM and

HRTEM images that a hollow CoO particle is created after oxidation at 200°C for 10 minutes.

Figure 13. HRTEM images from single domain ε phase Co nanocrystals (Coε-ORG) produced from organometallic

synthesis 8nm. (a) Before oxidation, single crystalline nanocrystal displaying the lattice plane from ε-Co. (b) After

oxidation at 200˚C for 10 min. Hollow particle showing the lattice planes for the CoO shell.

On increasing the temperature to 260°C (with 10 min exposure), hollow particles are obtained in the

TEM grid, like after the oxidation at 200°C and this can be seen in Figure 12g and 12h. The size of the

nanoparticle is 9.3 ± 0.69 nm, with a 7.4% distribution (inset Figure 12g), while the void has a

diameter of 2.5 ± 0.66 nm, with a distribution of 26.7%.This information was used to calculate the

shell thickness to 3.4 nm. This is slightly larger than the oxide thickness after oxidation at the lower

temperature (3.1 nm). The corresponding ED pattern (Figure 12i) reveals a change in the structure

after the change in oxidation temperature as the oxide shell is now built up by the spinal structured

Co3O4 instead of the cubic CoO. The distance between the lattice planes is larger for the spinal

structures and this explains the thicker shell after oxidation at the higher temperature.

The ε phase of Co is not as stable as the two other structures. It is more easily oxidized by the electron

beam in the TEM and it is likely that is also oxidizes easier during the oxidation treatment. The

formation of the hollow structure leaves no trace of a core as the 8 nm samples for both CoHCP-RM and

Copoly-ORG.

18

4.2. Size dependency

The results from the three sizes 4, 6 and 8 nm are presented for polycrystalline nanocrystals produced

from reverse micelle approach (Copoly-RM) and single domain hcp nanocrystals produced by annealing

Co nanoparticles from reverse micelle approach (CoHCP-RM). Both samples, in the three sizes, are

presented after the two oxidation treatments 200°C for 10 minutes and 260°C for 10 minutes.

4.2.1. Polycrystalline nanocrystals produced from reverse micelles (Copoly-RM)

A comparison between the three different sizes, 4, 6 and 8 nm, of Copoly-RM nanocrystals after

oxidation at 200°C is presented in Figure 14. The 4 nm particles have become fully oxidized CoO

(Figure 14a), the 6 nm particles have formed both fully oxidized CoO particles and core/shell Copoly-

RM/CoO with a small core (Figure 14b), while the 8 nm particles almost only display a core/shell

Copoly-RM/CoO structure (Figure 14c). The larger the particle is before oxidation, the larger the

resulting core becomes after oxidation. A more highly ordered area promotes the formation of a core

shell particle for a larger particle size, but as the particle size decreases all nanocrystals show complete

oxidation independently of the degree of order of the assembly. The crystalline structure of the CoO,

in any nanocrystal size, is similar to that of the original fcc nanoparticles with a rather low

crystallinity. The HRTEM images from the 4 and 8 nm sample are presented in the same figure

(Figure 14d and 14e). From the images, the smaller nanoparticles can be detected as fully oxidized

CoO with a very low crystallinity and the 8 nm nanoparticles as core/shell nanoparticles with a

polycrystalline CoO shell and a core with very low crystallinity.

Figure 14. Polycrystalline nanocrystals produced from reverse micelles (Copoly-RM) after oxidation at 200°C for 10 min.

TEM from three different sizes and HRTEM from the sizes 4 and 8 nm, (a) TEM 4 nm, (b) TEM 6 nm, (c) TEM 8 nm,

(d) HRTEM 4 nm, (e) HRTEM 8 nm.

The same comparison as described directly above is performed on the three sizes of Copoly-RM

nanocrystals after the other investigated oxidation treatment at 260°C for 10 minutes and is presented

in Figure 15. Both the TEM images and the corresponding ED pattern for the three samples are

showed. The 4 nm nanoparticles have coalesced and also became fully oxidized which can be seen in

19

Figure 15a. The corresponding ED pattern confirms a CoO structure, but as the native Copoly-RM had

very low crystallinity, the formed oxide has the same feature which can be seen by the unclear ED

pattern. The 6 nm nanocrystals have also formed fully oxidized structures (Figure 15c) but the

corresponding ED pattern (Figure 15d) reveals an uncertain oxidation state and it cannot be concluded

if the oxide is the cubic CoO or the spinal Co3O4. One reason for this behavior can be that the 6 nm

nanoparticle is in its transition state between the two oxidation states, but for a more elaborated

explanation an investigation of HRTEM images would have to be made. The 8 nm nanoparticles show

a mixture between fully oxidized and core/shell Co/Co3O4 nanoparticles (Figure 15e and 15f), the

nanoparticles are slightly coalesced but it is less severe than for the smaller sizes.

Figure 15. Polycrystalline nanocrystals produced from reverse micelles (Copoly-RM) after oxidation at 260°C for 10

min. TEM images and ED patterns from three different sizes, (a) TEM 4 nm, (b) ED 4 nm, (c) TEM 6 nm, (d) ED 6

nm, (e) TEM 8 nm, (f) ED 8 nm.

The oxidation behavior for Copoly-RM has a clear size dependency. After both oxidation temperatures

(200°C and 260°C) the 8 nm particles generate core/shell particles, however the oxide structure

differs. When the particle size is decreased to 6 nm or below, the oxidation results in a fully oxidized

structure, independently of the oxidation temperature. An interesting size dependency occurs after the

oxidation at 260°C where the size of the nanoparticles affects the obtained oxide structure. The larger

8 nm nanocrystals results in the spinal Co3O4 and the smaller 4 nm nanocrystal in the cubic CoO

structure. The 6 nm nanoparticle obtains a structure which cannot be decided to be one of the

described oxides after measuring the ED pattern, and a possible solution for this could be that the 6 nm

size is in its transition state in between the two possible oxide structures.

20

4.2.2. Single domain hcp nanocrystals produced by annealing Co nanoparticles from reverse micelle approach (CoHCP-RM)

A comparison between CoHCP-RM nanocrystals differing by their average size before oxidation (4nm,

6nm and 8 nm) and oxidized in similar experimental conditions as above (200°C; 10mn) was

performed. The resulting TEM images displays that the 4 nm CoHCP-RM nanocrystals (Figure 16a)

obtains a hollow single crystalline CoO structure with the HRTEM image (Figure 16d) showing the

CoO (200) lattice plane.

On increasing the CoHCP-RM nanocrystal size to 6 nm, hollow nanocrystals with both CoO single

domain and poly crystalline phases are observed (Figure 16b, 16e and 16f). Figure 16e shows the

formation of single domain hollow CoO nanocrystals with presence of the (111) planes, whereas

Figure 16f shows a hollow structure with a poly crystalline CoO lattice where the CoO (200) lattice

planes are oriented in different directions. The hollow void for both 4 and 6 nm is smaller than the

original Co nanoparticle indicating that the diffusion mechanism is not only outward diffusion of

cobalt, as this would have generated a void with equal size of the original particle. The smaller central

void can be due to some inward transportation of oxygen or due to inward relaxation of the void

through collection of the vacancies at the interface of cobalt and cobalt oxide. (8)

As already mentioned in chapter 4.1.2, 8 nm CoHCP-RM nanocrystals are characterized by core/shell

Co/CoO structure with a large core, indicating that oxidation at 200°C for 10 min is not enough to

make a nanoparticle of that size into a hollow structure.

Figure 16. TEM and HRTEM images from the three different sizes of single domain hcp nanocrystals produced by

annealing Co nanoparticles from reverse micelle approach (CoHCP-RM) after oxidation at 200°C for 10 min. (a) TEM 4

nm nanocrystal, (b) TEM 6 nm nanocrystal, (c) TEM 8 nm nanocrystal, (d) HRTEM 4 nm, single crystalline hollow

CoO particle, (e) HRTEM 6 nm, Single crystalline hollow CoO particle , (f) HRTEM 6 nm, polycrystalline hollow

CoO particle, (g) HRTEM 8 nm, core/shell Co/CoO with large single crystalline core.

After increasing the oxidation temperature to 260°C for 10 minutes the particles of all the sizes have

started to coalescence. The TEM image from the 4 nm nanoparticle (Figure 17a) shows fully oxidized

nanocrystals. The corresponding ED pattern (Figure 17b) reveals that the oxide has the cubic CoO

structure. The 6 nm nanoparticles (Figure 17c and 17d) looks fully oxidized at the first look, but as the

TEM image is more carefully studies some small hollow structures can be found. According to the ED

pattern the oxide structure is in this case also the cubic CoO. The TEM image and ED pattern for the

8 nm nanoparticles (Figure 17e and 17f) presents a hollow Co3O4 with a small void. Smaller particles

have a higher surface to volume ratio and thereby a higher surface energy, this mean that a hollow

21

particle is less stable the smaller the particle is and could explain why the 8 nm nanoparticles result in

a hollow particle while the 6 nm nanoparticles rarely presents a void and the 4 nm nanoparticles are all

fully oxidized.

Figure 17. TEM images from three different sizes of single domain hcp nanocrystals produced by annealing Co

nanoparticles from reverse micelle approach (CoHCP-RM) after oxidation at 260°C for 10 min. TEM images and ED

patterns from the three different sizes. (a) TEM 4 nm, (b) ED 4 nm (c) TEM 6 nm, (d) ED 6 nm, (e) TEM 8 nm, (f) ED

8 nm.

A clear size dependency can be seen among the CoHCP-RM nanoparticles after both oxidation

temperatures. The lower temperature (200°C) results in the same oxide structure (CoO) as the

oxidation for all samples at 200°C for 10 min for all sizes however the structure of the resulting

nanoparticle differs. The 4 nm nanocrystal obtains a hollow single crystalline structure while the

polycrystalline structure becomes more common for the 6 nm sample. The 8 nm nanocrystal does not

form a hollow nanoparticle at all but a core/shell CoHCP-RM/CoO instead. The higher oxidation

temperature (260°C) results in a different size dependent behavior. The larger the nanoparticle is the

larger the internal central void becomes. For the 6 nm nanocrystals a void can be seen in some cases

while the 4 nm nanocrystals presents a fully oxidized structure. Apart from this, the oxide structure

clearly differs as well. For the two smaller sizes the oxide is built up by cubic CoO while the 8 nm

nanocrystals results in the spinal Co3O4.

22

5. Conclusion

The expected result of the investigation was to find a change within the oxidation of the various single

domains, between the single- and polycrystals and between the different sizes of the Co nanocrystals.

Poly- and single domain Co-nanoparticles have different oxidation behavior. Single crystalline Co-

nanoparticles have a higher tendency to form hollow nanoparticles than polycrystalline Co-

nanoparticles. This can be explained by the grain boundaries ability to annihilate the created vacancies

without forming an internal void and thereby preventing the formation of a hollow nanoparticle in a

polycrystalline structure. The grain boundaries also work as a fast diffusion track for oxygen and

thereby enable the inward diffusion of oxygen to be faster than the outward diffusion of Co which

hinders the formation of hollow nanocrystals. The concentration of grain boundaries are important for

this behavior as the sample most near to be amorphous, polycrystalline nanocrystals produced from

reverse micelles (Copoly-RM), never showed any sign of creating a hollow particle, but the other

polycrystalline sample, Polycrystalline fcc nanocrystals produced from organometallic synthesis

(Copoly-ORG), formed a yolk shell structure.

The difference in oxidation behavior is also clear between the two single crystalline and between the

two poly crystalline structures. As mentioned above the concentration of grain boundaries and thereby

the size of the fcc domains makes up a clear different between the two polycrystalline structures. The

difference within the Copoly-ORG sample and the different composition if the fcc domains can be the

reason for the two different results after the oxidation treatment at 200°C, the hollow CoO and the

yolk/shell Copoly-ORG/CoO. Regarding the single crystalline samples the hcp structure (CoHCP-RM) is

more stable than the ε structure (Coε-ORG). This means that the cobalt atoms in the ε-lattice has a higher

mobility when heated and therefor an easier way for diffusion. This enables the structure to be more

readily transformed into a hollow structure since the formation of a hollow structure requires the

outward diffusion of Co to be faster than the inward diffusion of oxygen.

Isolated particles have in general a fast inward diffusion of oxygen as there is no protection given by

surrounding particles. When particles are ordered in a 2D hexagonal network the neighboring particles

slows down the rate of the inward oxygen diffusion and thereby promoted the outwards diffusion of

Co. Isolated nanocrystals therefor have a higher probability of being fully oxidized than ordered

nanocrystals. The oxidation behavior for the different crystal structures differ, but for all samples

where isolated nanoparticles were investigated, the isolated particle was closer to being fully oxidized

than the ordered nanoparticles in the same sample.

The size of the nanocrystal as well as the oxidation temperature (with a fixed oxidation time of 10

min) has an effect on which oxide (cubic CoO or spinal Co3O4) is obtained after oxidation. Oxidation

at 200°C always generates the cubic oxide for the examined structures while the structure generated

after oxidation at 260°C depends on the size and nanocrystallinity of the nanoparticle. For the single

crystalline CoHCP-RM the spinal structure is obtained after oxidation of the 8 nm sample at the higher

temperature while the cubic oxide is obtained for the 4 and 6 nm samples after the same treatment.

The higher oxidation temperature (260°C) affects the behavior differently than the lower oxidation

temperature (200°C) for the smallest single crystalline CoHCP-RM nanoparticles. The inward relaxation

seems to be made easier by the higher temperature, as the 4 nm becomes a hollow CoO for oxidation

at 200°C while it becomes fully oxidized CoO after oxidation at 260°C.

23

5.1. Future work

Concluding the thesis will be a few suggestions for future work within the area. A more complete

HRTEM study would be of interest, but as each sample takes a long time, the HRTEM images for the

samples oxidized at 200°C for 10 minutes were prioritized during the project. However it would give a

more detailed comparison between the two oxidation temperatures if the results were obtained for the

samples oxidized at 260°C for 10 minutes as well.

24

References

1. J.R. Davis. ASM Speciality Handbook: Nickel, Cobalt, and Their Alloys. pp.349-406. ASM

international, (2000). ISBN 0-87170-685-7.

2. J.D. Donaldson, D. Beyersmann. Cobalt and Cobalt Compounds. pp. 429-464. Ullmann's

Encyclopedia of Industrial Chemistry, (2005). Vol. 9. DOI: 10.1002 14356007.

3. H.C.Zeng. Synthetic architechture of interior space for inorganic nanostructures. Journal of

Materials Chemistry. 16, (2006), 649-662.

4. Q.Zhang, J.Xie, Y.Yu, J.Yang, J.Y.Lee. Tuningt the Crystallinity of Au Nanoparticles. Small. 6,

(2010), 523-527. DOI: 10.1002/smll.200902033.

5. Y.Tang, M.Ouyang. Tailoring properties and functionalities of metal nanoparticles through

crystallinity engineering. Nature Materials. 6, (2007), 754-759. DOI: 10.1038/nmat1982.

6. S. Sun, C.B. Murray. Synthesis of monodisperse cobalt nanocrystals and their assembly into

magnetic superlattices (invited). Journal of Applied physics. 85, (1999), 4325 - 4330.

7. Z.Yang, I.Lisiecki, M.Walls, M.P.Pileni. Nanocrystallinity and the Ordering of Nanoparticles in

Two-Dimensional Superlattices: Controlled Formation of Either Core/Shell (Co/CoO) or Hollow CoO

Nanocrystals. ACS NANO. 7, (2013), 1342–1350. DOI: 10.1021/nn304922s.

8. Y.D.Yin, R.M.Rioux, C.K. Erdonmez, S.Hughes, G.A.Somorjai, A.P.Alivisatos. Formation of

hollow nanocrystals through the nanoscale kirkendall effect. Science. 304, (2004), 711-714.

9. F. Aldinger. Controlled porosity by an extreme Kirkendall effect. Acta Metallurgica. 22, (1947),

923-928.

10. L. Cademartiri, G.A. Ozin. Concepts of Nanochemistry. Weinheim : WILEY-VCH Verlag

GmbH & Co. KGaA, (2009). ISBN: 978 3 527 32597 9.

11. H.J. Fan, U. Gösele, M. Zacharias. Formation of Nanotubes and Hollow Nanoparticles Based on

Kirkendall and Diffusion Processes: A Review. Small. 3, (2007), 1660-1671.

12. V.A. de la Peña O’Shea et al. Electronic and magnetic structure of bulk cobalt: The α, β, and ε-

phases from density functional theory calculations. The Journal of Chemical Physics. 133, (2010),

024701(1)-(8).

13. M. Cavalier, M. Walls, I. Lisiecki, M. P. Pileni. How Can the Nanocrystallinity of 7 nm

Spherical Co Nanoparticles Dispersed in Solution Be Improved? Langmuir. 27, (2011), 5014–5020.

Avaliable from: dx.doi.org/10.1021/la1049887 .

14. I. Lisiecki, M. Walls, D. Parker, M.P. Pileni. 2D Self-Organisation of Core/Shell Cohcp/Co

Nanocrystals. Langmuir. 24, (2008), 4295 - 4299.

15. P.Patnaik. Handbook of Inorganic Chemical Compounds. s.l. : mcgraw-Hill Professional, (2002).

ISBN: 0070494398.

16. S.C.Petitto, E.M.Marsh,G.A.Carson, M.A.Langell. Cobalt oxide surface chemistry: The

interaction of CoO(1 0 0), Co3O4 (1 1 0), and Co3O4 (1 1 1) with oxygen and water. Journal of

Molecular Catalysis A: Chemical. 128, (2008), 49-58.

17. J. Shi. Steric Stabilization. The Ohio State University. [Online] 08 29, 2002. [Cited: 02 16, 2013.]

http://www.matsceng.ohio-state.edu/ims/LR_Stericstablization.pdf.

18. Sigma-Aldrich. Sigma-Aldrich: Oleic Acid. [Online] 2013. [Cited: 04 25, 2013.]

http://www.sigmaaldrich.com.

19. L.E. Smart, E.A. Moore. SOLID STATE CHEMISTRY An Introduction. Boca Raton : Taylor &

Francis group, LLC, (2005). ISBN: 978 0 7487 7516 3.

20. F. Krumeich. Electron Diffraction (ED). Electron Microscopy ETH Zürich. [Online] 05 14, 2012.

[Cited: 01 25, 2013.] http://www.microscopy.ethz.ch/ED-1.htm.